ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

Talhada's profile - activity

2016-09-08 06:45:18 -0500 received badge  Teacher (source)
2016-09-08 06:45:18 -0500 received badge  Necromancer (source)
2016-09-08 06:42:16 -0500 received badge  Nice Question (source)
2013-09-06 19:38:28 -0500 received badge  Taxonomist
2012-09-19 23:16:16 -0500 received badge  Famous Question (source)
2012-09-19 23:16:16 -0500 received badge  Popular Question (source)
2012-09-19 23:16:16 -0500 received badge  Notable Question (source)
2012-08-27 12:25:13 -0500 received badge  Notable Question (source)
2012-08-27 12:25:13 -0500 received badge  Famous Question (source)
2012-08-27 12:25:13 -0500 received badge  Popular Question (source)
2012-05-10 08:16:55 -0500 answered a question stereo_image_proc : Assertion `left_.fx() == right_.fx()'

Hello! I have the same error:

stereo_image_proc: /tmp/buildd/ros-electric-vision-opencv-1.6.13/debian/ros-electric-vision-opencv/opt/ros/electric/stacks/vision_opencv/image_geometry/src/stereo_camera_model.cpp:30: bool image_geometry::StereoCameraModel::fromCameraInfo(const sensor_msgs::CameraInfo&, const sensor_msgs::CameraInfo&): Assertion `left_.fy() == right_.fy()' failed. Aborted

Here is my node launch file:

<group ns="xb3_short"> <node name="stereo_image_proc" pkg="stereo_image_proc" type="stereo_image_proc" args="_approximate_sync:=True _prefilter_size:=255 _prefilter_cap:=63 _correlation_window_size:=31 _min_disparity:=0 _disparity_range:=32 _uniqueness_ratio:=10 _texture_threshold:=100a _speckle_size:=400 _speckle_range:=15 __name:=xb3_short_stereo_proc"/> </group>

I checked the timestamps for left and right camera info topics and they match. Any idea? Thanks.

2012-05-02 03:36:55 -0500 received badge  Student (source)
2012-05-02 03:03:31 -0500 received badge  Editor (source)
2012-05-02 03:02:32 -0500 asked a question Bumblebee xb3 and stereo_image_proc

Hello everybody!

I have a question about stereo_image_proc and camera info manager.

I am using the Bumblebee xb3 camera, wich have 3 lenses. I need to get two pairs of images from the camera. One pair with the right and the center lenses (namespace=xb3_short), and another with right and left lenses (namespace=xb3_wide).

In order to run stereo image proc i need this topics:

xb3_short/left/image_raw                                                            
xb3_short/left/camera_info                                                    
xb3_short/right/image_raw                                         
xb3_short/right/camera_info                                                      

xb3_wide/left/image_raw                                                              
xb3_wide/left/camera_info                                                          
xb3_wide/right/image_raw                                                        
xb3_wide/right/camera_info

But actually xb3_short/right/image_raw and xb3_wide/right/image_raw are the same images. This will result a extra space when i need to record bag files with the camera. Is there a way to record only 3 images instead of 4?

2012-05-02 02:31:05 -0500 answered a question BumblebeeXB3

Hello, yes i am using xb3 with ros. But the package i am using was made by our team and is still not 100% fine.

2012-04-19 09:58:36 -0500 received badge  Supporter (source)
2012-04-19 09:57:42 -0500 answered a question Unmet dependency problem while installing electric: "but it is not going to be installed"

Yeah, i had the same problem and this one fixes it... thanks a lot

2012-04-16 08:02:38 -0500 asked a question 3D volume reconstruction

Hello,

Well i have a little problem and need some tips to solve it.

So, i have a stereo camera point cloud data, and i need to cluster it. It is very simple cluster the data directly from 3d using euclidean cluster extraction. However that is not my goal. I have also a laser scan pointed on the same direction with the camera and the laser gives me more believable information about clusters, however it is only 2D. The idea here is clustering the 2D laser scan with euclidean cluster extraction and then find a way to seed these points on stereo camera point cloud and then search for the neighbor points to estimate the size of this clusters. I looked on octree and kdtree search methods, but it is not what i needed. the octree gives me all the points inside a given radius from a seed point, and what i want is to see how long the cluster goes with a given threshold. Confusing?

anybody know a way to do that?